Down Pattern

A high-performance crossword clue and answer lookup service built with Cloudflare Workers and Hono.

Overview

Down Pattern (AKA "across-word") provides fast crossword answer pattern matching and clue lookup. It uses a chunked indexing strategy to efficiently serve crossword data from Cloudflare R2 storage.

API live at https://downpattern.livediagonal.com if you would like to cheat on crosswords or whatever

Features

• Pattern Matching: Find crossword answers matching a pattern (e.g., A?PLE finds APPLE)
• Clue Lookup: Get clues for specific crossword answers
• High Performance: Chunked indexes for fast lookups without loading entire datasets
• Scalable: Built on Cloudflare Workers with R2 storage

API Endpoints

GET /answers?pattern={pattern}

Find answers matching a crossword pattern.

• Pattern format: Use ? for unknown letters (e.g., A?PLE, ??T, HELLO????)
• Returns: Array of matching answers with occurrence counts

Example:

curl "https://your-worker.your-subdomain.workers.dev/answers?pattern=A?PLE"

GET /clues?answer={answer}

Get clues for a specific crossword answer.

• Answer: The crossword answer to find clues for
• Returns: Array of clues (up to 10 random clues)

Example:

curl "https://your-worker.your-subdomain.workers.dev/clues?answer=APPLE"

Data Architecture

The service uses a chunked indexing strategy:

1. Source Data: Large TSV file with crossword answer-clue pairs
2. Chunked Processing: Data is split into manageable chunks based on answer patterns
3. R2 Storage: Chunks are uploaded to Cloudflare R2 for fast retrieval
4. Dynamic Loading: Only relevant chunks are loaded for each query

Development Setup

Prerequisites

• Node.js 18+
• npm or yarn
• Cloudflare account with R2 enabled

Installation

1. Clone the repository:

git clone <repository-url>
cd down-pattern

2. Install dependencies:

npm install

3. Set up your crossword data:
   • Place your compressed TSV file at resources/clues.tsv.xv
   • The file should contain tab-separated answer-clue pairs

Building and Deploying

1. Build chunked indexes:

npm run build-chunks

2. Upload chunks to R2:

npm run upload-chunks

3. Deploy to Cloudflare Workers:

npm run deploy

Development Commands

• npm run dev - Start local development server
• npm run build - Build the Worker
• npm run build-chunks - Build chunked indexes from source data
• npm run upload-chunks - Build and upload chunks to R2
• npm run deploy - Deploy to Cloudflare Workers

Configuration

Configure your Cloudflare R2 bucket in wrangler.toml:

[[r2_buckets]]
binding = "DATA_BUCKET"
bucket_name = "your-bucket-name"

File Structure

├── src/
│   ├── index.ts                 # Main Worker application
│   ├── chunked-data-loader.ts   # Chunked data loading logic
│   └── types.ts                 # TypeScript type definitions
├── scripts/
│   ├── build-chunked-index.js   # Build chunked indexes
│   └── upload-chunks.js         # Upload chunks to R2
├── resources/
│   └── clues.tsv.xv            # Compressed source data
└── chunked-indexes/            # Generated chunk files (gitignored)

Performance

• Cold start: ~100-200ms (loads only needed chunks)
• Warm requests: ~10-50ms
• Memory usage: Low (chunks loaded on-demand)
• Storage: Efficient chunked storage in R2

License

MIT